Building a Headphone Measurement Lab

[rhythmdevils]

Tyll, it's really cool to see measurements of the Grado HP2!  

   Unfortunately, the only graph I can really make sense of is the frequency response.  Can you explain some of the other ones in simple terms, or is there a place where you or someone else has done this already? 
 
thanks!
 
as far as frequency response, it looks like above lower bass response, the HP2 and LCD-2 are very similar.

 

[Sambones]

Dude, there weren't as many headphones in the CanJam 2010 results PDF as I expected to see, but maybe I'm underestimating how long it takes. It would have been nice to see a wider assortment of orthos...which is what I came here for.
 
And, I know that graphs don't tell everything, but what's up with the Qualia? The THD measurements seemed to be quite poor...am I reading it wrong?
 
And, wow, "astonishing" doesn't begin to describe the LCD-2.

 

[xnor]

Quote:

And, I know that graphs don't tell everything, but what's up with the Qualia?

It sucks, technically speaking.

 

[kwkarth]

Quote:

a quick way to skim a paper is to jump to the end (of part 2):
 
 

 
From the graphs the HD800 doesn't look too shabby either - may not need much eq

Shaping basic frequency response is one thing, but you can't 'EQ" away slowness.  It would defy the laws of physics.  Fortunately, the HD-800s should be relatively fast for dynamic drivers.
 

 

[kwkarth]

Quote:

Dude, there weren't as many headphones in the CanJam 2010 results PDF as I expected to see, but maybe I'm underestimating how long it takes. It would have been nice to see a wider assortment of orthos...which is what I came here for.
 
And, I know that graphs don't tell everything, but what's up with the Qualia? The THD measurements seemed to be quite poor...am I reading it wrong?
 
And, wow, "astonishing" doesn't begin to describe the LCD-2.

Well, for orthos, Tyll did the HE-5, 5LE, and 6, the Audeze LCD-2.  He did the Thunderpants (Fostex T50RP on steroids)  And last, but not least, he measured the ultimate can, duggey's orthdome, the first can to achieve fully 100%... harmonic distortion...
 
What else is there?
 

 

[Sambones]

Haha, yeah. When you've got 8 drivers in each ear, that's what you can expect to get. But, I didn't see the HE-5LE, I just saw the LCD1 and three LCD2's.

 

[xnor]

Quote:

Shaping basic frequency response is one thing, but you can't 'EQ" slowness.

That's what impulse responses are good for. Are we moving in circles? :-/
 
But FR is the most important thing anyway (to listeners). (There are papers on this as well)
Fast transient response, slowness, resolution, clarity etc. you name it - often just a result of certain FR characteristics. I'll try to find a link.
 
 
edit: btw, are pics available on thunderpants internals?

 

[Pyriel0]

I am not understanding the compensation on the frequency graphs. After the compensation, it should just be a flat line correct?

 

[rhythmdevils]

I took these pictures while I had them.  Not sure if that's what you're looking for
 
http://www.head-fi.org/forum/thread/482221/loaner-program-smeggy-made-fostex-t50rp-woodies-aka-thunderpants/15#post_6551615

 

[xnor]

Thanks, rhythmdevils. What's behind the plexiglas though?

 

[odigg]

Quote:

Shaping basic frequency response is one thing, but you can't 'EQ" away slowness.  It would defy the laws of physics.  Fortunately, the HD-800s should be relatively fast for dynamic drivers.
 

Could you define slowness for me.  I've seen it used in two ways, one of which is clearly wrong.
 
1.  Headphones that have a tipped up treble response and reduced bass as seen to be "fast."  Headphones that have a tipped up bass response and reduced treble as seen to be "slow."  This may be how people perceive headphones, but it's not an accurate measure of speed.  My experience with the LCD2 made me think it sounded a little slow, but that's only because I felt the treble was a little less than what I thought it need to be "flat."  Obviously the LCD2 is a fast headphone.
 
2.  You are talking about a headphone's quickness as far as responding to changes in the original signal from the source.  If a headphone really is slow, that means it's not going to be able to reproduce that signal and something is going to be added (e.g. distortion).
 
I've yet to see any headphone (above $20) that was so slow that it couldn't reproduce the signal that was being sent to it.  Other than the example of "perfect distortion" in Tyll's measurements, can you point out a headphone that is so slow it can't reproduce the signal being sent to it?  Keep in mind that the FR can be different.  We are just talking about speed.
 
As for the Qualia, a number of people at CanJam (me included) fell vindicated when we saw those measurements (right after Tyll measured) because our subjective evaluations matched what those measurements insinuate.
 
And people wonder why most every division of Sony is in the red!

 

[kwkarth]

Quote:

That's what impulse responses are good for. Are we moving in circles? :-/
 
But FR is the most important thing anyway (to listeners). (There are papers on this as well)
Fast transient response, slowness, resolution, clarity etc. you name it - often just a result of certain FR characteristics. I'll try to find a link.
 
edit: btw, are pics available on thunderpants internals?

I agree, FR is what people hear most keenly.  One cannot take a physically massive, slowly moving driver and EQ it to move faster.  The acceleration factor is a function of moving mass vs. motor strength.  A fast can like the Beyerdynamic T1 or Senn HD-800 or Audeze LCD-2 can be made to emulate a slower can, like the Beyer DT-250, but no amount of EQ can make a DT-250 accelerate like a T1. 
 

 

[xnor]

It's especially the sub-bass that will make it sound "slow", but this has nothing to do with the driver's "speed".
(A peak at mid/upper bass and rolled-off sub-bass, like many dynamic headphones have, is likely to sound "faster" than another headphone with flat response down to 20 Hz. Of course, because the drums don't stick out but instead lower bass will.)
 
What's even more interesting, however, is that I don't think that any song is produced for reproduction on transducers that are ruler flat down to 20 Hz. Hence, I would only use such a headphone for monitoring purposes or something like that. It just won't sound right when listening to music, there has to be some kind of roll-off. Look at specs of monitors that engineers use for mixing/mastering (an extreme example would be the venerable ns-10m that starts to roll-off at around 200 Hz)
 
 
edit: @kwkarth: true of course!

 

[kwkarth]

Quote:

Could you define slowness for me.  I've seen it used in two ways, one of which is clearly wrong.
 
1.  Headphones that have a tipped up treble response and reduced bass as seen to be "fast."  Headphones that have a tipped up bass response and reduced treble as seen to be "slow."  This may be how people perceive headphones, but it's not an accurate measure of speed.  My experience with the LCD2 made me think it sounded a little slow, but that's only because I felt the treble was a little less than what I thought it need to be "flat."  Obviously the LCD2 is a fast headphone.
 
2.  You are talking about a headphone's quickness as far as responding to changes in the original signal from the source.  If a headphone really is slow, that means it's not going to be able to reproduce that signal and something is going to be added (e.g. distortion).
 
I've yet to see any headphone (above $20) that was so slow that it couldn't reproduce the signal that was being sent to it.  Other than the example of "perfect distortion" in Tyll's measurements, can you point out a headphone that is so slow it can't reproduce the signal being sent to it?  Keep in mind that the FR can be different.  We are just talking about speed.
 
As for the Qualia, a number of people at CanJam (me included) fell vindicated when we saw those measurements (right after Tyll measured) because our subjective evaluations matched what those measurements insinuate.
 
And people wonder why most every division of Sony is in the red!

1.  You are correct.  FR countour is NO indication of driver quickness, and people often confuse one for the other..
2. correct.
 
One of the most telling of Tyll's measurements are the two square wave response graphs.  Compare the LCD-2 or even HD-800 square waves with those from the ATH-L3000 or the Koss KDE-250, Sony V900, or Sony Qualia.
 
These plots are very revealing of real driver quickness and ringing.  More later... I have to run a time sensitive errand... will pick this up when I get back...
TTYL

 

[xnor]

Quote:

One of the most telling of Tyll's measurements are the two square wave response graphs.
...
These plots are very revealing of real driver quickness and ringing.

Yeah, quickness, ringing, damping.
 
And, in my book, that's about it what should be read out of those square waves.